Old Lithuanian “žmuo”, “žmunį” against the background of the Proto-Baltic athematic nouns

ACTA LINGUISTICA LITHUANICA LIETUVIŲ KALBOTYROS KLAUSIMAI XLIII (2000) Wojciech SMOCZYŃSKI Uniwersytet Jagielloński, Kraków OLD LITHUANIAN ŽMUO, ŽMŪNI AGAINST THE BACKGROUND OF PROTO-BALTIC ATHEMATIC NOUNS* 1. Stlit. žmuõ ‘person; man’ appears in Dauksza’s “Postilla” as a relic word, retreating before the neologism žmógus/žmògiš ‘ditto’.1 The archaic character of žmuõ is evident in its defective paradigm. It consists of three forms: nom. sg., acc. sg., and nom. dualis; cf. Kudzinowski 1977: I, 472 ff.: nom. sg. žmuo 10x, zmuo 2x (alongside žmógus, žmogiš)3 acc. sg. žmūni 1x, žmūni 1x (alongside žmógų) nom. du. žmūne 1x, žmūne 1x (there are no dual forms from the stem žmogu-). 2. The oblique forms acc. sg. žmūni and nom. du. žmūne occur in Dauksza’s “Postilla” in the following sentences: Acc. sg. žmūni: Ko iiemus padėsti bediewes tiesos kurės įstatinėje už žmūni dėsimti grivnų, o už szunį dwidėszimti arba daugiesn’ 339₂₉; Calwinas... pasamdęs žmūni giwą 459₅. Nom. du.: du žmūne lipo Bažniczion 313₁₉, du žmūne lipo Bažniczion 314₁₆ 3. K. Būga believed that the marked letters <ú> and <û> in Dauksza’s “Postilla” should be assigned the value of long /ū/ and therefore proposed reading the acc. sg. as *žmūni and the nom. du. as *žmūne4. However, the diacritic letters <ú>, <û> in Dauksza can just as well be interpreted as marking accent on the short vowel /u/. This is supported, first, by the identical spelling <ú> in žmūni and in forms of the word ‘dog’, which is known with certainty to have had oblique cases with short u in Old Lithuanian; cf. acc. sg. šūni 339₂₉, dat. sg. šūni 116₁₅, nom. pl. šūnes 5x. Further cf. šūke 406₄₇ = šūke * This is an expanded version of the article “Old Lithuanian žmuo: Reconstruction of the Paradigm,” published in Collectanea linguistica in honorem Casimiri Polański, ed. M. Brzezina and H. Kurek, Kraków: Uniwersytet Jagielloński, 1999, pp. 219–229. 1 Obie te akcentuacje są poświadczone u Daukszy, zob. Kudzinowski 1977: I, 427 n. 2 Cf. Kudzinowski, l.c. 3 Cf. Mažvīds: βmū 5x, szmū 2x, βmu 1x, alongside βmagus 9x and βmogus (numerous occurrences). There are no oblique forms with the stem žmun-; see Urbas 1996: 441. 4 Būga 1921: 424 (1959: 336). See also 1959: 214, note 1. 18 “shell,” súkas’ 139₁ = sùkas (i) “turns,” púse 3x, pússe 2x = pusè “part, half.” Secondly, on the basis of spellings with <û>, such as, for example, sunùs 256₄₀ = sùnùs “son,” lapùmo 41₃₀ [alongside lapùmo 32₄₁] = lapùmo “pride, haughtiness,” or sùnkum as 151₄₀ = sùnkum as “heaviness,” one may also assign to the letter <û> in the acc. spelling žmùni the value of short /u/⁵. In what follows, I will therefore use the transcriptions acc. sg. žmùni and nom. du. žmùne. Regardless of the indications afforded by examining the marked vowel letters in Dauksza’s “Postilla,” the objection to the reading *žmùni will become clearer against the background of the discussion below of the Proto-Baltic structure of the paradigm in question. 4. The acc. sg. form žmùni and nom. du. form žmùne, alongside nom. sg. žmùō, show that we have before us a noun with the suffix -n-, which is inflected athematically, in a manner similar to šùō, šùni. The purpose of the following remarks is to determine how, alongside the nominative form žmùō, a weak stem of the shape žmun-. could have become established. 5. In light of research to date, it is considered certain that nom. žmùō represents the Proto-Baltic structure *žm-ōn-Ø, where žm- represents the root in the zero grade, while -ōn- is the suffix in the lengthened grade, characteristic of the Indo-European asigmatic nominative. The analysis *žm-ōn is supported by the parallel formation of Lithuanian nom. sg. šùō “dog.” Unlike žmùō, the noun šùō has an external cognate, namely Vedic nom. sg. ś(u)vā, and can therefore be derived with much greater probability from the Proto-Baltic form *śwōn < PIE *k’w-ōn-Ø⁶. Both *śwōn and *žmōn are suffixed formations in -n-, once characterized by suffixal ablaut -on- (acc., voc.⁷):: -n- (gen.):: -ōn- (nom.). 6. The identity in word-formation structure of the nominatives žmùō and šùō makes it probable that the inflection of these two nouns will also be parallel. In the historical period, their forms that can be compared with one another are as follows: nom. sg. šùō žmùō acc. sg. šùni žmùni gen. sg. šùnes (*žmùnes) 5 Similarly Fraenkel, LEW 1320. It is worth noting that Skardžius 1956: 163 objected to Būga’s reading on the cogent grounds that the ending -uo of nom. sg. never alternates with the -ūn- of the oblique cases. 6 The Lithuanian reflex presupposes dissimilative loss of *s before *ō, which later developed into Lithuanian uo. Alternatively, one could speak here of an occasional simplification of the initial cluster *św-. Cf. Lithuanian šešuras “husband’s father; wife’s father,” also “brother-in-law” in Old Lithuanian < *šwešuras < Proto-Baltic *swešura- < PIE *swek’uro- (Avestan xvasura-, Greek hekurós). 7 Cf. Vedic acc. śvánam, voc. śvan. 19 nom. pl. šùnes (*žmùnes) nom. du. šùne žmùne 7. Commentary on the paradigm šùo. 7.1. Lithuanian nom. sg. continues IE *k’w-ōn, cf. Vedic śvá, śúvá (voc. śvan), Armenian šown (š < *k’w), Greek kúōn (voc. kuón). This proto-form may reflect a transformation of the nominative *pk’w-ōn (earlier *pek’w-ōn?), which developed on the basis of the zero grade in the oblique cases, of the type gen. sg. *pk’un-és. Underlying this derivative in -on-/-n- probably lies the root noun *pek’u-/*pk’u- ‘cattle’⁸. Perhaps this refers to the ‘dog’ as ‘guardian of the cattle herd’. 7.2. Gen. sg. šùnes (today šuñs) corresponds in the ablaut of the suffix to Vedic śúnas (instead of *śúnás), Greek kunós and can therefore be regarded as continuing the IE genitive in *k’u-n-és/ós. 7.3. Vedic acc. sg. śvánam ‘dog’ derives, taking Brugmann’s law into account, from *śvá.nam < IE *k’w-ón-m, i.e. from a formation characteristic of a strong case. The suffix here is full-vocalic and accented. Given the correspondence of the two Lithuanian forms with their Vedic equivalents: nom. šùo like śvá, gen. šùnes like śúnas, one would expect Vedic accusative śvánam likewise to correspond in Lithuanian to *švānį < Proto-Baltic *šw-an-in. This, however, is not the case, since Lithuanian shows šù-n-į. The accusative stem šun- replaced *švan-, probably under the influence of the genitive form šùnes. The motivation for this replacement should be sought in the tendency to limit paradigmatic ablaut in the prehistoric developmental stage of Lithuanian. In this case, the Proto-Baltic three-way alternation *šwōn:: *šwan:: *šun gave way to a Lithuanian alternation with only two allomorphs, Lithuanian šùo:: šùn. 7.4. Dytematism of the type šùo-:: šùn- is in principle consistent with what is observed in other consonant-stem paradigms. As is well known, Lithuanian athematic inflection does not distinguish strong and weak cases, but is based on two stems: one is specific to the nominative, while the other characterizes the casus obliquus. The difference between the inflection šùo, šùni (žmùo, žmùni) on the one hand and the other consonant stems on the other lies only in the fact that the latter generally show the consonant suffix in the casus obliquus in its full form (-en-, -er-, -el-, -es-), rather than in its reduced form, as in šù-n-ės, šù-n-i. Compare, respectively: (A) Lithuanian stems in -en-: 1. vand-en- (no *vand-n-, *und-n-, *ud-n-9), nom. vand-ùo ‘water’; 2. akm-en- (no *akm-n-), nom. akm-ùo ‘stone’. 8 Zob. Osthoff 1901: 230. Ostatnio Hamp 1980; Schmitt-Brandt 1998: 147, 207. 9 Cf. lexicalized words (1) Latvian ùdens m. ‘water’ < *undenis), (2) Old Prussian unds m. Ench. < *undis alongside wundan EV. acc., read [undan], probably neuter. The Baltic forms presuppose metathesis of the consonant n in the inherited weak stem *ud-n- (full grade: nom. *wéd-ōr, Hittite ù-i-da-a-ar = /wedār/). Cf. Vedic abl.-gen. sg. ud-n-ás, loc. sg. udán, udáni. – Lithuanian dialect forms: unduõ (puntininkai), ondù ‘water’, ùndens ‘waters’ (Samogitia) do not contain the inherited und-, but are phonetically derived from vanduõ; see Vitkauskas 1986: 103 ff. 20 (B) Lithuanian stems in -er-: 1. mot-er- (no *mot-r-), nom. mótē ‘mother; woman’ (cf. gen. sg. Latin mātris, Greek mētrós, Skt. mātúr [< *meh₂-tr̥-s]); 2. dukt-er- (no *dukt-r-¹⁰), nom. duktē ‘daughter’; 3. ses-er- (no *ses-r-¹¹), nom. sesuō ‘sister’. (C) Proto-Lithuanian stem in -el-/-ol-: *āb-el-/*āb-ol- (no *āb-l-¹²), nom. ábuolas ‘apple’ (Latvian ābuõls), obelis alongside obalìs ‘apple tree’. (D) Lithuanian stem in -es-: 1. mén-es- (no *mén-s-, cf. Latin mēnsis, Greek mēn¹³), nom. mén-uo ‘moon; month’; 2. deb-es- < *nebh₂-es-¹⁴ (no *deb-s-), nom. debesìs, -iẽs ‘cloud’, gen. pl. Old Lithuanian debesų. The conditions under which stems with the suffix in full grade e spread in the oblique cases have not yet been investigated. It may be assumed that forms of the hysterodynamic type in the accusative sg.¹⁵, which bore the accent on the suffix, played an important role in this innovative process. 7.5. As for the inflection of the Lithuanian word ‘dog’, the generalization of the zero-grade stem šun- in the casus obliquus (eliminating *švan-) has a parallel in Greek. Cf. Lithuanian gen. šunès:: Greek kunós, Lithuanian acc. šùnį:: Greek kúna (nom. šuõ:: Greek kuōn). 7.6. Given the role of the genitive sg. as the form from which the weak stem spreads throughout the entire paradigm, it is worth citing as examples those Lithuanian and Slavic nouns that are known only in the weak form: (A) Zero grade of the root: 1. Lithuanian širdìs f. ‘heart’, acc. šìrdį, Proto-Slavic *sъrdь-ce ⇐ Proto-Slavic acc. *sъrd-in, a form modeled on gen. *sъrd-és < IE *ḱṛd-és (full grade: Lithuanian šérdį ‘tree core; bell heart’ ⇐ *ḱerd-, Homeric Greek kēr, Hittite kēr¹⁶); 2. Lithuanian dùrės pl. ‘door’ ⇐ gen. ¹⁰ But outside the paradigm there is a derivative that attests the inherited stem *duk̑tr-. This is Lithuanian pódukra ‘stepdaughter’ < *pā-duktr-ā. As for the simplex dukrà ‘daughter’, it is more likely a compound in relation to pódukra than a suffixal derivative, *duk̑tr-ā. ¹¹ Cf. Ved. gen. svásur (*swésr-s-), Goth. swistrs, as well as the fact that in Slavic a derivative *sesr-ā, OCS sestra “sister,” is a trace of *sesr-. ¹² The weak stem *āb-l- is attested by Slavic derivatives: OCS jābl-anъ, -i f. “apple tree” (< *ābl-ōn-) and *abъlъko “apple” (< *ābl-u-ko-). It is doubtful, however, that such a stem was also known in Prussian. Pruss. woble “apple” (Elbing Vocabulary), a hapax, alternates with the consistent *(v)ābel- in Grunau’s Glossary: wabelcke, wabelko, wobelke. Thus woble may be a secondary form, syncopated or Germanized in its spelling (on this, see Smoczyński 1986: 36 n.). ¹³ The stem *mēn-s- was continued in Slavic as well, although it underwent a series of transformations there: > 1. *mēns-i- ⇒ 2. *mēnsi-ko- > 3. [n metathesis] *měsinko- > 4. OCS měsečь. Stages 1–2 find parallels in the Slavic treatment of the IE stem *sh₂-ul- “sun”: ⇒ *sul-ni- ⇒ *sul-ni-ko- > OCS sъlnьce (cf. § 7.6. B). Erhart 1998 etymologizes měsečь differently. ¹⁴ Cf. Ved. nábh-as- n. “moisture; mist; cloud,” nábhasī dual “heaven and earth”; OCS nebo, nebese “heaven.” ¹⁵ For the terminology of inflectional types characterized by mobile accent and ablaut, see Schindler 1975, Rix 1976: 123. ¹⁶ The lengthened grade that characterized the nom. sg. accounts for the acute intonation of the Lithuanian stem šérd-; the acute in the weak stem šìrd-, on the other hand, is analogical to the nom. sg. 21 IE *dʰur-ós (full grade *dʰwor-, OCS dvorъ “place behind the door”¹⁷); 3. OCS krъvь “blood” ⇐ Proto-Slavic acc. *kruw-i-n¹⁸ < *kruH-in, a form modeled on the gen. *kruw-és < IE *kruh₂-és (full grade: Ved. kravi-s, Gk. kréas “dripping with blood, raw meat” < *kréwh₂-s-¹⁹; 4. OCS vьsb “village” ⇐ IE gen. *wik’-ós (nom. *woyk’-s, cf. Old Lithuanian vieš-pats “lord,” Old Prussian steimans buttas waispattin dat. pl. “den Hausfrawen, ladies of the house,” Gk. oíka-de “homeward”). (B) Zero grade both in the root and in the suffix: 1. OCS sъnъ “sleep” < *sopnъ ⇐ IE gen. *sup-n-és²⁰ (nom. *swép-ōr, Lat. sopor, -oris m. “deep sleep,” Lith. sãpnas “dream”); 2. OCS dьnь, dьn-e “day” ⇐ IE gen. *di-n-és (nom. *dy-ōn/*dy-ēn)²¹; 3. Proto-Slavic *sъl- from IE *sh₂-ul- “sun” (⇒ *sul-ni-²² ⇒ *sulni-ko- > OCS sъlnьce)²³; 4. East Lithuanian bruvís f., Church Slavonic brъvь ⇐ Proto-Slavic acc. *br-uw-in²⁴; 5. Proto-Lithuanian *dr-u- “tree” (cf. Ved. dros gen. sg. “wood” < *dr-ew-s, Gk. drûs f. “tree,” drutómos “woodcutter”) is the base of the vṛddhi derivative Lith. dervà f. “pinewood; resin extracted by burning wood” < *d-e-rw- + -ó- “made of wood, wooden”²⁵; 17 The form OCS dvьri pl. “door,” Russ. dverь is problematic, particularly against the background of the absence of a form *dōri, with the prevocalic allomorph *dur- as in Lith. dùres, secondarily dùrys “door.” The presence of a syllabic r after v (vьr) presupposes in any case the sequence *wr-C, not *wr-V; cf. Slavic četvŕt-ō:: Lith. ket-vìr-tas “fourth” (but ket-ur-i “four”). It may be supposed that the Slavic word for “door” is a derivative with a consonantal suffix, *dwir-yo- < *dʰur-yó-. Its structure is like that of Gk. thâirós m. “hinges on which a door turns” < *tʰwar-yó-. 18 Cf. the Lithuanian adjectival derivative krùvi-nas “bloody” = OCS krъvь-nъ “related by blood, kindred.” 19 The Lithuanian reflex of this prototype would be *kriáuš-. As for Lith. kraujas “blood,” owing to (a) the circumflex and (b) the diphthong au after a hard consonant, the structure of this formation remains puzzling. One may posit a thematic derivative *krewh₂-o- “flowing” > (laryngeal loss) *krewo- > *krowo- (cf. OCS novъ:: Gk. néos “new”) > Proto-Baltic *krava-, from which the collective kraujai was formed with the suffix -yo-? 20 Here and below, in reconstructing the genitive ending, I omit the ablaut variant *-ós. 21 The same weak stem is presupposed by the Lithuanian vṛddhi derivative dienà “day” < *deynā < *deyn-eh₂- ⇐ *d-e-y-n + -ó-, from *din-. Cf. the type diẽvas “god” < *d-e-yw-ó- “connected with the sky,” from *diw- ⇐ *dy-ew- “daylight sky.” See most recently Smoczyński 1998: 62 n. (with references). 22 A suffix apparently modeled on og-nъ “fire,” cf. Ved. ag-ni-. 23 The Proto-Slavic weak stem *sōl- < *sh₂-ul- is comparable, on the one hand, with Ved. gen. súras < *suh₂-és (this results from metathesis of *sh₂-ul-), and, on the other, with the strong stem *sh₂-wel- in Ved. svàr-/súvar- n. “sun.” Lith. sáulė, however, appears to be a vṛddhi-type derivative from the metathesized weak stem: *sh₂-ul- > *suh₂-l- ⇒ *s-e-wḱ-l-ó- adj. “solar” > Proto-Baltic *sāwla-. Apparently, a derivative in -iyā-, *sāwlyā, was formed from this adjective and eventually became substantivized. See most recently Bader 1995, who adopts the reconstruction *seh₂-l- for Lat. sōl, citing the development of mōtus < *mouH-to-. 24 Instead of *br-yaw-in < IE acc. *bʰr-éwH-m (cf. Lith. briau-nà “edge, boundary, ridge”), as a result of the leveling of the accusative form to the gen. *br-uw-és < IE *bʰr-uH-és, with secondary w, which filled the hiatus after the loss of the laryngeal (cf. Gk. ophrus, ophruós “eyebrow”). 25 Indo-European neuters with a vṛddhi structure, especially OCS drevo “tree; wood” < *denw-o- and Old English teoru “wood tar” < *terw-a-, suggest that Lithuanian feminine dervà may originally have been a collective noun, with -a from *-eh₂ (= nom. pl. n.). 22 6. Prussian pintis “road,” if it is a derivative from a genitive stem *pint-és < *pnt-h₂-és (cf. Vedic pathás)?26. (C) Zero grade of the root and full grade of the suffix: OCS imę, imene “name” < *jbmen- < *im-men-es ⇐ gen. *in-mén-s < IE *(H)nh₃-mén-s (cf. full grade in Latin nōmen, Sanskrit nāma < *(H)neh₃-mn; no Balto-Slavic *nōmin). Of this list, the stems šun- are chiefly suitable for structural comparison with *di-n- (OCS dъnъ) and *sup-n- (OCS sъnъ). 8. The paradigm of žmuō. 8.1. It is generally held that žmuō < *žm-ōn is related to the Indo-European word for “earth” and is a derivative meaning “earthly being, earth-dweller”27. Schindler 1967: 200 and 1975: 263 reconstructed the paradigm of the noun “earth” as follows: nom. *d(ʰ)égʰ-ōm cf. Hittite <te-e-kán> = /deg-an/ acc. *d(ʰ)égʰ-om-m̥ cf. Greek kʰtʰ-ón-a < **tʰkʰ-óm-am loc. *d(ʰ)gʰ-ém-Ø cf. Vedic ks-ám-i gen. *d(ʰ)gʰ-m-és cf. Vedic j-m-ás 8.2. In the Balto-Slavic languages, traces of only two forms of the above paradigm are attested, namely the singular locative and genitive. 8.2.1. The Proto-Baltic stem *žem- should be connected with the IE locative *d(ʰ)gʰ-ém > *gʰém; from it was derived the Balto-Slavic word for “earth”: Lithuanian žemė, OCS zemlja < *žem-(i)yā. Judging by the suffix, it was originally an adjective28, but the conditions of its substantivization have not yet been explained. In addition to the stem in -(i)yā-, Slavic also attests a derivative in -i-, *gʰem-i-. It became established, on the one hand, in the Russian adverbs na zemb “onto the earth,” o zemb “against the earth,” and, on the other, in the OCS adjective zemьnъ “earthly” < *žemi-na-, which contrasts 26Meanwhile, Slavic shows OCS pъtь, -i m. “road,” whose stem in -i- is connected with the resegmentation of the old acc. sg. form *panti-n ⇐ *pant-in < IE *pónt-h₂-m (Vedic pántham ⇐ *pántam). Traditionally, a Proto-Balto-Slavic paradigmatic alternation *pant-:: *pint- is assumed here. At this point, however, a certain reservation arises, since in the Prussian corpus, alongside <in>, pintis, spellings with <en> also occur, namely pentes and peentis “road” (in Grunau). These spellings with <en> suggest a phonetic derivation of pintis from pentis < *pantis (the vocalism of the Protoform in Prussian would thus agree with OCS pъtь). Such a derivation has a close parallel in the variation among forms of the Prussian participle *sant- “being,” whose Old Lithuanian counterpart is sant-, acc. sg. santį (cf. Latin sōns, sontis “guilty, deserving punishment” < IE *h₃s-on-t-). These forms are, on the one hand, -senti- (gen. sg. emprikisentismu “of the present”), and, on the other, -sint- (nom. sg. emprijkisins “present”). The Prussian change *anC > enC > inC is most probably explained by interference from Low German phonetics; cf., for example, Middle Low German sinte “holy” < sente < sante (Lasch 1914, § 139). For other examples of secondary inC in Old Prussian, see Smoczyński 2000: 19n. 27 Zob. Skardžius 1956: 163, Fraenkel, LEW 1320. 28Cf. the Vedic adjective kṣām-ya- “located on the earth, earthly, humanus,” opposed to divyá- “heavenly, diuinus” (Fraenkel 1959: 13). 23 from OCS zemljьnъ from zemlja29. – Thus, the Proto-Baltic *žem- consisted of ž as a fragment of the original root in the zero grade30 and em as the original suffix in the full grade. 8.2.2. The Baltic neo-root žm- is associated with the Indo-European genitive *d⁽ʰ⁾gʰ-m-és > *gʰm-és (two zero grades); from it derives the word that interests us, *žm-on- ‘earthly being, earth-dweller’ > ‘human being’31. The corresponding neo-root is also known in Slavic, but there it appears only in the name for ‘snake’ and the mythical ‘dragon’: OCS zmijь m. ‘drákōn, óphis’; zmija f. ‘óphis’; Old Russian zmii, gen. zmija; Russian zmei, zméja; Serbo-Croatian zmâj, zmâja, etc., all from Proto-Slavic *zmьb, gen. *zmbja, an adjective meaning ‘crawling on the ground’. The pre-Slavic form *žm-iyo- shows that this is an adjective meaning ‘earthly, terrestrial’. Originally it occurred in the syntagma *zmbьžь ‘earth snake’ (the semantic specification based on the opposition to ‘water snake’) and became substantivized following the ellipsis of the noun žь32. The absence in Lithuanian of the corresponding forms *žm-ijas, *žm-ijā is noteworthy (Būga 1959: 376). 8.3. Turning now to the reconstruction of the pre-Baltic inflection of the word *gʰm-on- > *žm-an-, it should immediately be noted that reconstructing a prehistoric inventory of case forms encounters two difficulties. The first is the lack in the Lithuanian evidence of an attestation for an athematic genitive singular form corresponding to žmuõ. The second difficulty is the presence of the oblique stem žmun- (žmùni, žmùne). This stem is usually segmented as žm-un-, but this cannot be justified on historical-phonetic grounds, since the structure *gʰm-n-C is excluded in the singular paradigm. I shall begin with those details that at present seem fairly clear. 8.4. The prototype of the nominative singular *gʰm-ōn is certainly clear; Old Lithuanian žmuõ derives from it through regular phonetic development. In reconstructing the suffix *ōn in the lengthened grade, in addition to the comparative data (see § 5), we also have some intrabaltic evidence. This consists of Lithuanian suffixes with the structure -uon-i- (e.g. gėluonis alongside gėlùo), -uon-e- (mómuonė alongside momuõ), -uon-a- (širšuõnas alongside širšuõ), -uon-ia- (ėduonia)), used mainly, though not exclusively, in denominal derivation; see Būga 1959: 333 ff., 368 ff. 8.4.1. It should be assumed that the suffixes mentioned arose through the thematization of old consonant stems in -n-, by attaching the paradigmatic features -i-, -e-, -a-, -ia- to the nominative singular form and by immobilizing 29 See Fraenkel 1959: 13 on this. 30 If the full grade had continued in Balto-Slavic, the nominative singular would have taken the form Lithuanian *dežuo and Slavic *deža. 31 Cf. the zero grade in Greek epi-kʰtʰón-ios ‘living on earth’ (perhaps also in neo-kʰm-ós ‘new, unexpected’). – An independent derivative from the strong stem of the noun ‘earth’ is represented by Sanskrit hemō ‘homo’ (cf. némō ‘nobody’), accusative hem-ōn-em ‘hominem’ < *gʰém-ōn. 32 See Vaillant 1950: 75; Fraenkel 1959: 4. 24 of the ablaut paradigm, which was a consequence of thematization. The question of how the mechanism of this thematization should be reconstructed nevertheless remains open. In my view, the sequence uon in the thematized nom. sg. in -uon-V- can be viewed as an amalgam that, from a historical point of view, is divisible into two segments. The first segment is uo as the ending of the athematic nom. form -uo < *-ōn (loss of *-n before a pause), and the second is n as an element of the suffix -en-/-an- in oblique cases with vocalic endings (gen. *-en-es, acc. *-en-in or *-an-in). The introduction of uo before n took place in the process of eliminating paradigmatic ablaut -uo (nom.):: -en-/-an- (obl. case), by generalizing uo. When an original paradigm, such as *širšuo:: gen. *širš-en-es, acc. *širš-en-i33 acquired a new shape, *širšuo:: gen. *širš-uon-es, acc. *širš-uon-i, the possibility naturally arose of abstracting the neostem *širšuon- from it and subsequently using it as a base for thematic derivatives; compare širšuon-as ‘hornet; wasp’ alongside širšuon-e, Old Lithuanian širšuonis, same meaning (LKŽ XIV 908). In turn, the emergence of back-formations of type (1) širš-e, širš-a (LKŽ XIV 906) ⇐ širšuonė, (2) širš-as (Chylinski, ibid.) ⇐ širšuonas provided the basis for resegmenting širšuonas ⇒ širš-uonas (now motivated by širšas), etc., and thereby for the establishment of -uonas, -uone, etc., as new nominal suffixes. They joined the ranks of the long-vowel suffixes -ūnas, -ynas, -ėnas, -onas. Evidence of this process is the fact that, alongside širšuonas, younger variants širšonas, širšūnas ‘hornet’34 appeared. Still later, by means of substitutive derivation -uon-:: -uol-, variants such as širšuolis, širšuolė. were formed. 8.4.2. In the case of an isolated form such as Old Lithuanian pirmuon-es nom. pl. f. SD ‘ancestors; elements, origins’, one may posit a stem in the form of nom. sg. *pirm-uo albo *pir-muo. The old acc. sg. of *pirmuo may have had the form *pirm-an-in, as indicated by the thematized forms known from the Prussian ‘Enchiridion’; compare nom. sg. pirm-onn-is ‘first’, dat.-acc. pirm-ann-in, pirm-ann-ien. 8.4.3. Just as stems in -n- were transformed into athematics in -l-. Of the derivative óbuol-as (Latvian ābuōls) ‘apple’, one may say that its relation to nom. *ob-uo < *āb-ōl:: obl. *ab-el- is analogous to the relation of širšuon-as to nom. širš-uō < *širš-ōn:: obl. *širš-en-. From the fact that alongside obelīs f. ‘apple tree’, a form with the suffix *-ol-, obalis f. of the same meaning (LKŽ VIII 985) is also attested, one may infer a double form of the Proto-Lithuanian acc. sg. and reconstruct, on the one hand, *āb-el-in > óbelį, and on the other, *āb-al-in > óbalį (compare nom. pl. óbelės, gen. pl. obelų). Given the o-grade in nom. sg. *āb-ōl, perhaps the older form should be considered 33 Compare Vedic gen. sg. śṛṣ-n-ás [nom. śíras ‘head’] < *k’r̥h₂-s-n-ós, stem *k’erh₂s-en-. For Slavic, compare Serbo-Croatian sršljen, Old Church Slavonic sršēń ‘hornet’, Proto-Slavic stem *sьrš-en-. 34 Against this background, I consider the presence of Lithuanian ablaut uo:: ūn alongside uo:: o, suggested by K. Būga’s lexical comparisons (1959: 331 ff.), less probable. 25 consider *āb-el-in to be the acc. sg., while *āb-al-in < *āb-ol-im should be regarded as an analogical form based on the nom. sg. (?). 8.4.4. Finally, it is worth noting that the Proto-Baltic stem in -t-, *nepuo, *nepátin, *neptés, was also thematized. Old Lithuanian nepuotis ‘grandson’35 is based on the oblique case *nepuot-i-, which became established in place of *nepat-i-, resulting from the resegmentation of the old acc. *nepát-in < Indo-European *nep-ót-m36. The sequence uo is again explained by the replacement of a with uo in the accusative suffix under the influence of nom. *nepuo < *nep-ōt or *nepuos < *nep-ōt-s (compare Vedic nāpát, Latin nepōs). A similar innovation is seen in Latin acc. nep-ōt-em (instead of *nep-ot-em), and even in gen. nep-ōt-is (instead of *neptis < *nep-t-es, compare Vedic náptur37). In both cases, the aim is to eliminate allomorphism resulting from inherited suffixal ablaut *ōt:: *ot:: *t. 8.5. The reconstruction of the acc. sg. as *gʰm-on-m, Proto-Lithuanian *žm-an-in38 is justified from a comparative perspective by the correlation nom. *-ōn-Ø:: acc. *-on-m; compare Old Attic hemo:: hemōnem ‘human being’, Greek téktōn:: téktōna ‘carpenter’, Vedic tákṣā:: tákṣāṇam, same meaning, Greek ákmōn:: ákmona ‘anvil’39. By contrast, there is no comparative support for a possible proto-form *gʰm-n-m, which would have two zero grades (as in the weak case). Certain internal evidence also supports *žm-an-in. 8.5.1. First, compare accusatives of the type *pirm-an-in to *pirmuo and Lithuanian dialectal óbalį to *obuo, mentioned above in § 8.4.2–3. Second, the existence in Lithuanian of the word žmónės f. ‘people’ < *žman-ē-, which functions as the nom. pl. of žmuo, seems to support the reconstruction of an acc. sg. in the form *žman-in. If this word is interpreted as a derivative in the lengthened grade, its stem must be reconstructed as *žman-. For the lengthened grade in o, compare Lithuanian žólė ‘grass’ from žalias ‘green’, mõlė ‘a period of grain milling’ from malù, málti ‘to grind’40. Less probable seems 35 Dat. pl. nepuotims (i-stem), acc. pl. nepuočius (yo-stem). 36 Alongside the masculine nepuotis, -iēs ‘grandson’, there was a feminine i-stem, nepuotis (Margarita Theologica). A derivative in -ē, nepuotė ‘granddaughter’ (Bretkun), arose alongside it; see LKŽ VIII 670. 37 The derivatives Old Lithuanian nept-ė ‘granddaughter’ and nept-is ‘grandson’ (LKŽ VIII 670) were formed from the Proto-Baltic genitive *nep-t-és. In time, these too were replaced by nepuotis and nepuotė, respectively. From other languages, compare Old Church Slavonic netii, Serbian-Church Slavonic netjb ‘cousin’ < *nept-iyo-, Vedic naptiya- ‘belonging to the progeny’, Avestan nauua.naptiia- n. ‘relatives of the ninth degree of kinship’, Vedic napti, Latin neptis, -is f. ‘granddaughter’. 38 Compare Stang 1966: 227. 39 Compare the correlation in r-stems nom. *-ēr-Ø:: acc. *-er-m, e.g. Greek anēr:: anéra ‘man’, mētēr:: mētera ‘mother’ (Lithuanian mótė:: móterį, Old Church Slavonic mati:: materb). 40 Do stopnia wzdłużonego *žmān- sprowadził žmón(ēs) już Peters 1980: 189, choć nie stwierdził wyraźnie, że *žman- jako podstawa tego derywatu należało do paradygmatu rzeczownika „człowiek“ i pojawiało się w formach acc. sg. i voc. sg. – Inaczej Fraenkel 1959: 12, który przyjmował, że pierwotny nom. pl. miał postać *žmuon-es (pluralizacja formy nom. sg. przez -es) i następnie uległ wykolejeniu w žmón-es (forma gwarowa z Twerecza), mianowicie na tle tendencji do 26 the supposition that the formation *žmān-ē f. is a collective formed from the vocative singular form *žman “O human!” < *gʰm-on-φ. Compare, on the one hand, vocatives such as Vedic śvan, Greek kúōn, and, on the other, Lithuanian vocatives such as dialectal sēser “O sister!”, dùkter “O daughter!”. 8.5.2. Traditionally, Balticists also identify the lengthened stem *žmān- (Lith. žmon-) in Old Prussian. They assume that Proto-Baltic *žmān- changed here through labialization to *[zmuːn-]. This pronunciation is supposed to be attested by forms with <smūn-> in the “Enchiridion,” such as smūnets once, “Mensch, human” (alongside smunent- 11 times), steison smūni once, gen. pl. “der Person[en], of the person[s]” (from denasalization of the so-called asigmatic gen. pl. *smūn-in), the denominative smūnint once, “ehren, honor” (alongside smunin- twice), and en smūnin once, dat. sg. “in Ehren, in honor.” In my view, this hypothesis is too weakly grounded. Leaving aside for the moment the hapaxes smūni, smūnin, what is evident above all is that, among 12 occurrences of the word smunent- “human,” it is written with a macron in only one (!) instance: smūnets 63₁₉. It is noteworthy that this spelling lacks the n component of the suffix. One is led to suspect that smūnets is a misprint for <smunēts> (the macron being misplaced), to be read smune(n)ts. If, then, the macron here was an abbreviation for n rather than a sign of post-accentual length, there is no philological basis for interpreting smunent- (thus 11 times) as [zmuːnent-]⁴¹. 8.5.2.1. As for the hapaxes smūni, smūnin, which evidently stand in an etymological relationship with smunent-, it must be stated that the mere presence of the macron in these spellings is not in itself sufficient to warrant reading <ū> as [uː]. It should be borne in mind that the “Enchiridion” contains several dozen examples showing a macron over vowels that are undoubtedly short⁴². This kind of spelling suggests that the macron is used primarily as an accent mark (an indicator of mobile word accent), whereas its connection with the quantity of a post-accentual vowel is secondary (cf. Smoczyński 1990). Here are instances of the spelling <ū> in words which, by comparison with their Lithuanian equivalents, may be interpreted as containing short [u]: 1. poklūsnai once, ‘gehorsam’ (alongside poklusm° 9 times), read [poklūs-]. the replacement of old uo by o in certain nouns; compare, for example, slōgas “burden” instead of slūogas (⇐ slēgti “to weigh down, press, suffocate”). – Zinkevičius 1980: 248 offers yet another explanation: žmōn-es instead of *žmun-es, with o supposedly generalized from the word žmōną “woman; wife.” Could the author really think that the vowel o in žmōną could have a different origin from the o in žmōnes (-es)? 41 This is how, for example, Mažiulis reads it, PEŽ IV 135, deriving it from Proto-Baltic *žmānent- “žmogelis, little man.” 42 Cf. 1. [i]: polinka ‘remains’:: Lithuanian dialectal palìnka, senìnka ‘gathers’:: Lithuanian dialectal surènka, kakìnti ‘to grasp’:: Lithuanian kákinti, alkìns ‘sober’:: Old Church Slavonic alčьnъ “hungry,” newìns ‘the ninth’:: Lithuanian devìntas, dessìntas ‘the tenth’:: Lithuanian dešìmtas, imt ‘to take’:: Lithuanian ìnti; 2. [ē]: gēhrt ‘to speak’, preigērb ‘to speak before’:: Lithuanian geibti, mērgan ‘maid’:: Lithuanian mergà, acc. mergą, sacramentan ‘sacrament’:: Lithuanian sakramentas, 3. [ā]: ānters ‘other, second’:: Lithuanian antras, rankans ‘hands’:: Lithuanian acc. pl. rankàs, perbānda ‘tries’:: Lithuanian bandó, bandýti. 27 Compare Old Lithuanian and Lithuanian dialectal paklùsnas m., paklusnà f. ‘the same’. 2. aulauūsins once, acc. pl. ‘the dead’, read [aulavúsans], an alternative spelling to aulauwussens I₉₅. Compare Lithuanian dialectal pasilióvusius ‘the deceased’. 3. pickùls ‘devil’, read [pikúls], cf. Lithuanian pikùlas, same meaning. In light of these facts, we must allow for the possibility that the spelling <smūni> may conceal a short-vowel stem [zmún-], that is, the same as the one represented by the spelling of the word smunents- “human.” This leads to the conclusion that the Prussian language may have had the stem [zmun-], which was, in origin, identical to the Old Lithuanian stem žmun-, acc. žmùni. All this leads to rejecting the hypothesis that isolated spellings such as smūn- could contain [muː] deriving from optional labialization of *[maː] and thus be etymologically related to the innovative Lithuanian form žmōnes. 8.5.3. Z. Gołąb (1998: 69) has recently invoked the Proto-Baltic acc. *žmanin in connection with a new attempt to etymologize the Slavic word mòžb “man.” He believes that the stem *žman- < IE *gʰmon- was shared by Baltic and Slavic, except that in Slavic it was transformed by means of the suffix -g- (the parallel would be Proto-Slavic *lěmǫgъ beside Lithuanian liemuõ, liemẽn-s). An accusative form based on such a stem would have been *žmangim ⇒ *žmang-i-m, yielding the Slavic i-stem *žmǫžь. The historical form mòžb would have arisen through dissimilatory loss of the first ž and the word’s shift into the -yo-stems. Gołąb’s proposal compares favorably with earlier etymologies of this word (*manu-gyo-, *man-g-yo-), among other reasons because it posits the presence of the suffix -g- also in the Lithuanian neologism žmogùs43. But it should be noted that the issue here depends on explaining the semantic function in which suffixation with *-g- or *-gʰ- was used in Balto-Slavic derivation. Unfortunately, this question has not yet been investigated. 8.6. Having discussed the issues relating to the Proto-Baltic acc. *žmanin, we now turn to the original genitive of the word “human.” The form of this case can only be inferred indirectly, namely by reference to the parallelism between the declensions of the nouns “dog” and “human” (see § 6). Comparing nom. *kʷw-ōn:: gen. *kʷ-u-n-és, one may suppose that alongside nom. *gʰm-ōn a genitive of the form *gʰm-n-és = [gʰmnés] would have been possible. In this formation, then, the suffix appears as a consonantal n, as a result of which the root m is realized in its syllabic allophone [m̥]44. In Baltic, as usual, we must take into account the possibility of two different vocalizations of the sonant. The reflex *[gʰmnés] could theoretically have taken two forms: either with imC, i.e. *žim-n-es, 43 Proponowany przez Gołąba 1998: 72 prototyp *žmō(n)g- jest chyba bez paraleli fonotaktycznej. Pomysł Otrębskiego (1950: 85), by w žmogùs upatrywać złożenie *žmān-magus (por. goc. magus „chłopiec“), które z czasem zostało przekształcone «par une contamination», nie znalazł oddźwięku u innych badaczy. Inne próby referuje Fraenkel, LEW 1318. 44 Compare cases of sonant vocalization before the affix -n- in other languages: Greek dám-nā-mi < *dm̥-né-h₂-; Latin iungunt < *yu-n-g- from *ywg-/*yewg-. 28 or from *umC, i.e., *žum-n-es. Support for this hypothesis may be found in the alternation im:: um < *[m] in such Lithuanian forms as (1) rìmtas alongside dialectal rùmtas “serious, dependable” (from remiùos, remtìs “to lean on, rely on something”) or (2) timpstù “I stretch” alongside ítumpas “impetus, swing” (from tempiù “I tighten”), cf. LKŽ IV 269⁴⁵. As will become clear from the following discussion, it will be more reasonable to assume a reflex of the genitive in the form Proto-Baltic *žum-n-es (§ 8.9). 8.7. The Lithuanian oblique case with the stem žmun- raises the question of how to interpret the segment un following the root žm-. If this un were a reflex of syllabic n, an anteconsonantal context would be required, *gʰm-n-C = [gʰm-n-C]⁴⁶. Such a context would be possible only in the plural inflection, namely before consonantal endings, as in loc. pl. -su (*žm-un-su “among people”) or instr. pl. -mis (*su žm-un-mis “with people”)⁴⁷. The fact is, however, that the Lithuanian record contains no plural forms of žmuõ whatsoever. This, of course, makes it impossible to verify the hypothesis proposed. Added to this is the fact that in Dauksza’s “Postylla” žmuõ is not accompanied by a plural form *žmunes, but by the suppletive form žmonės “people” (§ 8.5). As for the dual du žmùne, it is impossible to determine whether this form was created from the singular stem *žumn- (see the gen. *žumnes above), which was subsequently assimilated to the anteconsonantal plural stem *žmun-C, or whether it arose in some other way. 8.8. If we accept that žmun- did indeed arise phonetically before the consonantal plural endings, its presence before the vocalic endings (acc. sg. žmùn-i, nom. du. žmùn-e) will have to be explained by analogical leveling. In the singular, the generalization of žmun- would have had to lead to the elimination of the irregular genitive form *žum-n-es (the anomalous relation žm-uo:: *žum-n), as well as the accusative form *žm-an-i < *gʰm-on-m (cf. *švanį < *kʷ-on-m in § 7.3), which differed from the other forms in the ablaut grade of its suffix. 8.9. A weakness of the scenario described in §§ 8.7–8.8 is that we find no parallel among Lithuanian inflectional patterns for the analogous spread of the anteconsonantal plural stem to the oblique singular cases. In these circumstances, the following hypothesis suggests itself: ⁴⁵ For the position after ž, the uR/iR alternation may be illustrated by examples such as Latvian zułtis “bile,” Lithuanian tułžis, same meaning (instead of *žułtis < Proto-Baltic *žuł-ti- as against Proto-Slavic *zъłьcь “bile,” Old Russian zъłьcь, Russian želčь, Old Church Slavonic zъłьcь < *žil-k- from Indo-European *gʰlH-C (full grade in Lithuanian želti “to turn green,” Old Church Slavonic zelenъ “green”). ⁴⁶ I do not understand why Fraenkel, LEW 1320, writes the reconstruction in the form *gʰm-dn-. The “schwa” sign has no morphological justification here, since -n- represents reduction to -on-, not to a suffix containing a laryngeal (*-Hₒn- or *-gₒn-). ⁴⁷ Cf. the occasional Old Lithuanian instr. pl. forms of the type akmenis < *akmen-mis and seser-mis (Zinkevičius 1980: 245). 29 (A) The form of the stem žmun- is nevertheless genetically connected with the singular paradigm, in particular with the inherited genitive singular form. (B) The phonologically regular reflex of gen. *žum̥n-ès < *gʰm-n-és (§ 8.6) underwent a kind of metathesis, becoming *žmun-ès. A parallel for the displacement of the vowel u according to the formula umn > mun is the change ulž > luž in the Lithuanian name for “spleen”: blu̇žnīs f., acc. blu̇žnį apparently derives from (1) *plužnį48 < (2) *pulžnį < (3) Proto-Baltic *pulž-n-in < (4) Indo-European dialectal * (s)plgʰ-n-m. Cf. Old Indic plihán- “spleen,” Greek splēn, splēnós, Latin liēn < *lihēn49. (C) This was probably not a purely phonetic change50. Since its effect was to regularize the initial cluster žm- across all inflected forms, it should rather be assumed to have been a morphonological regularization brought about under pressure from the two forms beginning with žm-, namely nom. žmuõ < *gʰm-ōn and acc. *žmànį < *gʰm-on-m51. The consequence of changing *žum̥nès to *žmunès was the elimination of the anomalous root allomorphism *žum-:: žn-. 9. After the “metathetic” genitive *žmunes had become established, a further innovation followed, this time involving ablaut. It consisted in replacing the acc. form *žmanin with *žmunin. The allomorphism žmōn:: žman:: žmun was thus reduced to two members, žmōn:: žmun. This happened in the same way as in the paradigm of the noun “dog,” where šūni (with šun as in gen. šunès) replaced *švàni < *švanin (cf. § 7.3), with the result that the series švōn:: švan:: šun was reduced to švōn:: šun. As a result, the paradigm of the noun žmuõ became a two-stem paradigm, like that of the typical Lithuanian consonant declension (cf. § 7.4). 10. What has been said may be summarized in the following developmental schema. The sign “>” indicates a phonetic change, while the sign “⇒” indicates a morphological change. IE Proto-Baltic Proto-Lithuanian Lithuanian nom. *gʰm-ōn-Ø > *žm-ōn-Ø > *žm-ōn-Ø > žmuõ-Ø acc. *gʰm-ōn-m > *žm-an-in > *žm-an-in ⇒ žmùn-i gen. *gʰm-n-és > *žum-n-es ⇒ *žm-un-es > *žmun-ès 48 bl- is explained by the sonorization of pl-, like that seen in the word blusà “flea” < *plusà, from Old Indic plúši, Old Lithuanian. Further compare the cluster lb in Lithuanian gulbė “female swan” with Slavic *kʰolpъ, Kashubian kiełp “swan”. 49 The Slavic equivalent looks different again: Old Church Slavonic slězena < *selžena < *spelgʰ-en- (loss of p in initial position). For Balto-Slavic one might posit the ablaut paradigm *(s)pelž-(e)n-:: *(s)pulž-(e)n-. The Baltic continuation would involve generalization of the weak stem *(s)pulž-n-. 50 As regards the phonotactic aspect of this change, it should be taken into account that Lithuanian has not a single word beginning with *žum- or *žun-. It does, however, have words with the initial cluster žm- (cf. the family žmogùs, as well as žmegìns, žmùlys) and with the initial cluster žn-: žnáibyti, žnýbti, žnýplès. 51 This formulation draws on a suggestion made by Prof. Vytautas Ambrazas during the discussion of my paper at the conference “Lietuvių kalba: tyrėjai ir tyrimai. Kazimiero Jauniaus 150-osioms ir 90-osioms mirties metinėms paminėti”, Vilnius, 22–23 October 1998. 30 The development of the Lithuanian noun “dog” paradigm can be presented in a similar way: IE Proto-Baltic Proto-Lithuanian Lithuanian nom. *k’w-ōn-∅ > *šw-ōn-∅ > *šw-ōn-∅ > šuō-∅ acc. *k’w-ōn-m > *šw-an-in > *šw-an-in ⇒ šun-i gen. *k’u-n-és > *šu-n-es > *šu-n-es > šun-ès A comparison of the two schemata shows that the regularization of the “dog” paradigm is achieved through the simple generalization of the inherited gen. sg. stem šun- to the acc. sg. Meanwhile, in the “human” paradigm we are dealing with a sequence of two changes: first, the transformation of the inherited genitive *žum-nes into *žmunes, then the generalization of the innovative stem žmun- to the acc. sg. Only as a result of both these changes did the striking similarity between the forms žmuō and šuō emerge in the historical period⁵². Cf. in particular the alternation uo:: un, which is not repeated in any other Lithuanian noun. 11. It has been attempted here to show that the declension type to which the consonant-stem “human” belonged presupposed, on the one hand, the presence of a gen. sg. in the Proto-language form *žum-n-és, and, on the other, the presence of an acc. sg. in the form *žm-án-in. The replacement of acc. *žmánin by *žmānin (Lith. žmùni) could not have taken place without the presence of a formal model in the shape of the genitive *žmunes. But it must be remembered that such a model appeared only when the inherited gen. *žum-n-és, under the pressure of forms with initial žm- (*žmōn, *žmānin), assumed the new form *žmunés. Because of the key role played in the development of the paradigm reconstructed above (§ 10) by the genitive form *žum-n-és, we cannot agree with the account to date, which explained the emergence of the acc. sg. form žmùni by means of the analogical proportion: nom. šuō:: acc. šùni = nom. žmuō:: acc. X (X = žmùni)⁵³. Such a formulation of the proportion tacitly assumes the semantic dominance of the noun “dog” over the noun “human,” which is already unlikely for semantic reasons. The main flaw in the traditional explanation, however, lies elsewhere. It does not address the question of where acc. šùni came from (it is, after all, not an inherited form); moreover, it suggests as though acc. žmùni had no morphological predecessor in the strong form *žmanin. BIBLIOGRAFIA Bader F. 1995: Autour de lat. sōl. – Linguistica Baltica 4, 265–276. Būga K. 1921: Priesagos -inas ir dvibalsio uo kilmė. – Lietuvos mokykla – 10–11, 420–457. Przedruk w Būga 1959: 331–376. ⁵² Być może na tej podstawie Peters 1980: 188 doszedł do przekonania, że *gʰm-ōn (žmuō), musi być neologizmem wzorowanym na *k’wōn (šuō). ⁵³ Zob. Specht 1932: 214, 226; Specht 1934. Stang 1966: 227 ujmuje rzecz ostrożniej: „Vielleicht war die ursprünglich Flexion žmuo, *žmani, *žmunes. Später wurde die schwundstufige Form des Stammes verallgemeinert, wie in šuō, šùni, šùnis.“ 31 Būga K. 1959: Rinktiniai raštai 2, sudarė Z. Zinkevičius, Vilnius. Erhart A. 1998: Der indogermanische Mondname. – Linguistica Baltica 7, 63–69. Fraenkel E. 1959: Zur indoeuropäischen Stammbildung und Flexion. – Lingua Posnaniensis 7, 1–24. Fraenkel E. LEW = Lituaisches etymologisches Wörterbuch, Heidelberg, 1955–1965. Gołąb Z. 1998: Lat. homo ~ Slavic možь: is there any etymological connection between them? – Studia Indogermanica Lodziensia 1, 67–73. Hamp E. P. 1980: IE *()k’uon ‘dog’. – Indogermanische Forschungen 85, 35–42. Kudzinowski Cz. 1977: Indeks-słownik do «Daukšos Postilė», I, Poznań. Lasch A. 1914: Mittelniederdeutsche Grammatik, Halle a.S. LKŽ = Lietuvių kalbos žodynas, Vilnius 1941–. Mažiulis V. PEŽ = Prūsų kalbos etimologijos žodynas, 1–4, Vilnius, 1988–1997. Osthoff H. 1901: Etymologische parerga, Leipzig. Otrębski J. 1950: Miscellanées onomastiques. – Lingua Posnaniensis 2, 70–98. Peters M. 1980: Untersuchungen zur Vertretung der indogermanischen Laryngale im Griechischen, Wien. Rix H. 1976: Historische Grammatik des Griechischen. Laut- und Formenlehre, Darmstadt. Schindler J. 1975: Zum Ablaut der neutralen s-Stämme des Indogermanischen. – Flexion und Wortbildung, hg. von H. Rix, Wiesbaden, 259–267. Schmitt-Brandt R. 1998: Einführung in die Indogermanistik, Tübingen u. Basel. (Uni-Taschenbücher, 1506). Skardžius P. 1956: Alte Wurzelnomina im Litauischen. – Indogermanische Forschungen 62, 156–166. Smoczyński W. 1986: Uwagi o słowniku bałtycko-słowiańskim. – Język i jego odmiany w aspekcie porównawczym, red. J. Majowa (Prace Slawistyczne PAN 53), Wrocław, 17–45. Smoczyński W. 1990: Zur altpreußischen Akzentographie. – Rivista di Linguistica 2:2, 177–199. Smoczyński W. 1998: Il ruolo della lingua lituana per la linguistica indoeuropea. – Ponto-Baltica 7, 1997[98], 53–82. Smoczyński W. 2000: Untersuchungen zum deutschen Lehngut im Altpreussischen, Kraków. Specht F. 1932: Beiträge zur griechischen Grammatik. – Kuhns Zeitschrift 59, 214, 226. Specht F. 1934: Lit. žmuo. – Kuhns Zeitschrift 61, 68. Stang Chr. S. 1966: Vergleichende Grammatik der baltischen Sprachen, Oslo-Bergen-Tromsö. Urbas D. 1996: Martyno Mažvydo raštų žodynas, Vilnius. Vaillant A. 1950: Grammaire comparée des langues slaves, Lyon-Paris. Vitkauskas V. 1986: Jest’ li takaja forma: unaduo ‘voda’? – Balto-slavjanskie issledovanija 1984, 103–106. Zinkevičius Z. 1980: Lietuvių kalbos istorinė gramatika 1, Vilnius. SENOSIOS LIETUVIŲ KALBOS ŽMUO, ŽMUNĮ ATSIŽVELGIANT Į ATEMATINIŲ DAIKTAVARDŽIŲ PARADIGMĄ Santrauka Daukšos Postilėje acc. sg. žmunį ir nom. du. žmune turi kirčio ženklą ant trumpojo skiemens. Kiti netiesioginiai linksniai nepaliudyti. Pagrindinis klausimas: kaip atsirado kamienas žmun-, kurio negalima paaiškinti pagal kitas priebalsinio -en- kamieno formas. Remiantis formų šuo:: šunės santykiu, kur genityvo forma atitinka paveldėtą modelį (ide. *k’u-n-es), reikėtų tikėtis daiktavardžio žmuo genityvo *žumnes. Tai būtų praformos *g’hm-n-es atitikmuo. Priesaga su neskiemeniu -n- būtų sąlygojusi skiemeninę šaknies baigmens formą, kurios prūsųlietuviškas refleksas būtų -um-. Tačiau žm(uo) alomorfą žum(n)- pakeitė žmu(n)-. Iš pažiūros tai panašu į metatezę: *žum(n)- > 32 *žmu(n)-, bet jos priežastis neaiški. Čia šakininis morfas žm- galėjo būti pagal nominatyvą apibendrintas kitiems linksniams (pakitimą galėjo skatinti ir ta aplinkybė, kad lietuvių kalboje nėra žodžių, prasidedančių *žūm-). Pagal ved. švanam < *kʷ-on-m pavyzdį reikėtų laukti akuzatyvo formos pral. *šv-an-in ‘šunį’. Priesagos nykstamąjį laipsnį čia reikėtų aiškinti genityvo formos įtaka: šun(i) pakeitė *švan(i) pagal šun(es) pavyzdį. Panašiu būdu pirminė forma *žman(i) galėjo prisiderinti prie genityvo žmun(es) ir virsti paliudyta forma žmuni. Aiškinant genityvo lemiamą vaidmenį daiktavardžio žmuo paradigmos kitimui, remiamasi ir lietuvių bei slavų kalbų silpnųjų atematinių kamienų daiktavardžiais, pvz., lie. širdìs, dùrys, sl. vъsь. Atkreiptas dėmesys ir į bendresnį dalyką: trinariai alomorfai (pvz., *švōn: *švan: *šun-) paprastai keičiami dvinar ais (pvz., šuo: šun-). Pateikiama ir daugiau „bitematizmo“ polinkį rodančių lietuvių kalbos linksniavimo pavyzdžių. OLD LITHUANIAN ŽMUO, ŽMUNĮ AGAINST THE BACKGROUND OF THE PROTO-BALTIC ATHEMATIC NOUNS Summary In Daukša’s Postilė, the forms acc. sg. žmunì and nom. du. žmune have a stress mark placed over a short syllable. No other oblique cases are attested. The principal question that rises here is how to account for the stem žmun-, which cannot be explained in the framework of the consonantal declension with a stem in -en-. If we take as our point of departure the relationship between the forms šuo:: šun es, where the genitive corresponds to the inherited model (PIE *kʷ-u-n-es), we are entitled to expect alongside žmuo a genitive *žumnes. This would be a reflex of *gʰm̥-n-es. The suffix, in its non-syllabic variety -n-, would have induced a syllabic realisation of the root-final -m-, the regular Proto-Lithuanian reflex of which would be -um-. The allomorphic relationship of žum(n)- to žm(uo) was, however, eliminated in favour of žmun-. On the face of it, this looks like a metathesis *žumn- > *žmun-. But the motive behind this change remains unclear. A possible explanation could be that the root morph žm- spread from the nominative to the oblique cases (the fact that Lithuanian has no words beginning with žūm- could somehow have contributed to this). In the accusative we would expect, on the basis of Vedic śvānam < *kʷ-on-m, a Proto-Lithuanian reflex *šv-an-in ‘dog’. The generalisation of the zero grade of the suffix should be attributed to the influence of the genitive: šun(i) replaced *švan(i) under the influence of šun(es). By analogy, we could assume an original *žman(i) to have been assimilated to the genitive žmun(es) to yield the attested žmunì. In order to explain the crucial role played by the genitive singular in this remake of the paradigm of žmuo, the parallel of the Lithuanian and/or Slavonic nouns built on weak grade athematic stems, such as širdìs, dùrys, vъsь, is invoked. Attention is also drawn to an observation of a more general nature, viz. that a ternary allomorphic series (e.g., *švōn: *švan-: *šun-) is usually replaced with a binary one (e.g., šuo: šun-). Other instances of this ‘bithematism’ in Lithuanian declension are cited in support of this. Wojciech SMOCZYŃSKI Uniwersytet Jagielloński Katedra Językoznawstwa Ogólnego i Indoeuropejskiego al. Mickiewicza 9/11, PL-31-120 Kraków, Polska El. paštas [email protected] Gauta 1999 06 16 33